ML20011A466

Evaluation of Allowable Tensile Stresses Normal to Bed Joint.

ML20011A466
Person / Time
Site:	Point Beach
Issue date:	09/30/1981
From:	COMPUTECH ENGINEERING SERVICES, INC.
To:
Shared Package
ML19312F019	List: ML19312F018 ML20011A461 ML20011A463 ML20011A466 ML20011A469 ML20011A476 ML20011A483 ML20011A484 ML20011A488
References
IEB-80-11, R553.06, TAC-42896, TAC-42897, NUDOCS 8110130446
Download: ML20011A466 (24)

Text

EVALUATION OF ALLOWABLE TENSILE STRESSES NORMAL TO THE BED JOINT Prepared for Point Beach Nuclear Power Plant, Units 1 and 2 WISCONSIN ELECTRIC POWER COMPANY Milwaukee, Wisconsin Prepared by COMPUTECH ENGINEERING SERVICES, INC.

Berkeley, California September,1981 REPORT NO. R553.06 i

8110130446 811007 ,

PDRADOCK05000g ,

G 1

TABLE OF CONTENTS 1 INTRODUCTION . . . . .................... ..... 1 2 OVE RVIEW OF TEST P ROG RAMS . . . . . . . . . . . . . . . . . . . . . . . .. . 2 2.1 APPLICABILITY OF TEST RESULTS ........... ...... ... 2 3 EVALUATION OF MONOTONIC TEST RESULTS . ................ . 4

3.1 DESCRIPTION

OF STATISTICAL ANALYSES . . . . . . . . .. .... .. 5 3.2 RESULTS OF STATISTICAL ANALYSES . . . . . . .... .. 7 3.2.1 Sample Statistics ..... ................ .... 7 3.2.2 Confidence Intervals on the Population Mean . . . ..... . .. 7 3.2.3 Discussion of Normal vs Gamma Distribution . . . . . . . . . . . . 8 3.2.4 95% Confidence Levels Corresponding to le .20 .and 30' Levels . 9 3.2.5 Safety Factors Based on the Mean . . . . . . . . . . . . . . . ... 9 3.2.6 Probabihties of Exceedance . .................... 9 3.3 DISCUSSION OF RESULTS . . . . . . . . . ............. ... 10 4 EVALUATION OF SHAKE TABLE TEST RESULTS . . . . . .......... . 12 4.1 RESULTS OF TESTS ON HOUSE 1 . . . . . . . . . . . . . . . . . . . . . 12 4.2 RESULTS OF TESTS ON HOUSE 2 . . . . . . . . . . . . . . . . . . . . . 13 4.3 RESULTS OF TESTS ON HOUSE 4 . . . . . . . . . . . . . . . . . . . . . 13 4.4 DISCUSSION OF TEST RESULTS . . . . . . . . . . . . . . . . . . . . . . . 14 5 SPECIFIC RESPONSE TO ACTION ITEM 4 . . . . . . . . . . . . . . . . . . . . 15 6 CONCLUSIONS . . . .............................,. . 18 7 REFERENCES . . . ................... . . . . 19

1 INTRODUCTION ine Nuclear Regulatory Commission (NRC) staff on June 9-11, 1981, reviewed the criteria and calculations performed on IE Bulletin 80-11

• Masonry Wall Design" for tne Point Beach Nuclear Power Plant. Action item 4 resulting from that meeting stated that with regard to allowable tensile stresses normal to the bed joint the licensee shall provide the following:

A. Applicability of NCMA and other test data to the Point Beach masonry walls, with particular emphasis on the static nature of the test as compared to the dynamic / seismic loading being evaluated.

B. Discussion of the results obtained from dynamic tests conducted at Berkeley and discuss their nonapplicability with respect to the bed joint tensile stresses.

C. Discussion and justification for higher r owable tensile stresses for SSE than those allowed for OBE. It is the NRC position that such an increase should not exceed 33-1/3% of the OBE values.

D. Technical discussion demonstrating that high safety margins in wall capacity exist even after first significant cracking has occurred due to out-of-plane shaking in light of the tests conducted at Berkeley.

E. Statistical summary of the static monotonic test data as described in item 6.a.

This report provides the information required on the five items above and is subdivided as follows. Section 2 provides an overview and discusses the applicability of the test programs from which the test data is obtained. Sections 3 and 4 present an evaluation of the static and shake table test data respectively.

Section 5 provides the specific response to each of the five items and Section 6 presents the conclusions.

1

2 OVERVIEW OF TEST PROGRAMS The results of seven different test programs regarding the tensilo strength of mortar normal to the bed joint are evaluated in this report. Six of the test programs, given in references 1 through 6. involved static. monotonic load tests. and one, given in references 7. 8 and 9. invohed shaking table tefsts.

The six static test programs provided results for 81 unreinforced test specimens.

involving four different mortar types, namely. M. S. N and O as specified by proportion in ASTM C270. Also varying between the six static test programs was die way in which the walls were loaded. Some tests were carried out using a uniform pressure (air bag) loading. some used concentrated center point loading, and others were performed with concentrated loads at the qua-ter points of the wall. The uniform load produces a parabolic moment distribution over the height of the wall, tne central loading condition produces a symmetric triangular distrioution with a maximum at midspan, and the quarter point loading produces a region of constant moment over half the neight of the wall, in one series of experiments (3) the walls were tested after orily 15 days of curing.

The shake table test program provided results on four unreinforced concrete block walls constructed with type S mortar. The major drawback with the results of this test program is the difficulty in assessing the cracked state of the unreinforced walls before the tests began. Each wall was constructed away froen the shake table, and was moved into place prior to testing. Except for visible cracks. the extent of microcracking that occurred during the moving operation could not be readily determined.

2.1 APPUCABILITY OF TEST RESULTS The results of the 81 tests performed in the six static test programs, in our opinion, are applicable in determining the tensile strength normal to the bed joints for seismic loads. for the following reasons:

1. An unreinforced masonry wall responds elastically to seismic loads provided it is not cracked. This was demonstrated in some of the shaking table tests.

2. There are no test results available indicating that dynamic

! loading reduces the tensile strength normal to the bed joint. In fact the only test cata available for any type of cyclic loading on masonry structural elements indicates that the in-plane shear strength of masonry shear walls tested pseudostatically is 8-23% less than that of a 3 cps equivalent dynamic test (Reference 10).

3. Cyclic or shake table tests are essential to determine the post-cracked or inelastic performance of structural elements. However, they are not essential to determine ihe ultimate or cracking strength of structural elements.

2

4 Points 1. 2 and 3 above indicate that the uniform or point load tests are reasonable methods to determine the cracking or tensile strength of an unreinforced masonry wall subjected to out-of-plane loads.

The shake table tests cannot be relied upon M determine the cracking or tensile strength of the uncracked, unreinforced walls because it cannot De ascertained that the walls were not Cracked before the tests began.

In f act major Cracking was observed in several of the test specimens prior to testing (See Figs. 5.1(a) and 5.12(a) of Reference 7 and Figs. 4.2(a) and 4.19(a) of Reference 8.) Other microcracking may have occurred but was not detected prior to testing. Therefore, the shake table test 'esults cannot ce used to datermine the tensile strength normal to the 'Oed joint.

However they are valuable in assessing the post-cracked performance of unreinforced out-of-plane walls subjected to seismic loads.

r 3

3 EVALUATION OF MONOTONIC TEST RESULTS The results from six different monotonic test programs (1.2.3.4.5.6) on the tensile strength of mortar normal to the bed joint form the basis of the statistical analysis presented in this section. In total. data from 81 tests were available, involving four different mortar types, namely types M. S. N and O. Only the results of tests with type N mortar, as specified by proportion in ASTM C270.

are sised herein as this was the mortar type specified for the Point Beach Nut %ar Plant. Tests reported in (1) and. (6) contain no data for tjpe N mortar.

and thus have no further part in this study.

The followirg table indicates the large variabliity between the remaining tests on type N mortar. It was necessary to make the modifications indicated to the data from (3) and (5) in order that all section moduli were based on the net mortar bedded area.

Reference No. of Loading Comments tests 2 8 uniform Section Modulus based on mortar bedded area 3 14 1/4 point Tensile strengtn based on gross area. Values are multiplied by 1.4 to compare to net area. Tests were carried out after only 15 days o' curing.

4 3 1/4 point Section modulus based on mortar bedded area (83 sq.in/ft) 5 18 center point Tensile strength based on gross area. Values are multiplied by 1.14 to compare to net area.

Tensile strengtn normal to the bed joint is influenced by several variables, perhaps the single most important of which is the mortar cube strength. The 43 samples with type N mortar (8 uniform load. 17 quarter point load and 18 center point load) cover a wide range of cube strengins from 610 psi on the low end to 2500 psi on the high end. The effect of this variable is taken into consideration when evaluating the tensile strengths applicable at Point Beach.

In two separately reported studies the effect of the loading condition (quarter point loading versus uniform loading) on the apparent tensile strength was evaluateo. The first stody by Morik (11) prcouced a theoretical analysis indicating that quarter point load tests would give tensile strengths apparently lower than uniform load tests, the actual difference being a function of the coefficient of variation of the mortar strength. For typical values of this coefficient of variation for type N mortar. the analysis indicated that point load tests would give tensile 4

strengths approximately 10 % lower than would uniform load tests. The paper then went on to compara tensile strerigth results from the two dit.sent kinds of loading. for the case of brick walls and found the experimental ratio between the mean strength from uniform load tests and it.e mean strength fr.?m quarter point load tests to be 1.97. Although the paper did not specifically recommend the adoption of this tactor of 1.97 to relate quarter point load data to uniform loao data, the factor seems to have been used blindly in the past for this purpose. The second study (12) again looked at experimental data, and came up with a factor of 1.99 for concrete masonry walls.

it is instructive at this point to examine the reason why quarter point loading produces lower apparent tensile strengths than does uniform loading, in the case of a uniform load, a parabolic moment distribution results, subjecting one joint to the maximum moment. However, for quarter point loading, fully one half of the wall or typically some 6 to 7 joints are subjected to the same maximum moment. Given the inherent variabihty in mortar strength, failure will occur in this case at the weakest joint of the 6 or 7, which may be well away from the center of the span. The same joint, under a parabolic moment distribution. may not fall since it would be subjected to lesser moment. Thus the apparent strength of the mortar normal to the bed joint will, on the average, be lower for the quarter point load cases. Strengths from uniform load tests and center point load tests would be expected to be similar. since in both cases only one joint is subjected to the maximum moment.

While references (11) and (12) have indicated use of " correlation factors' of 1.97 and 1.99 respectively to relate quarter point load data to uniform load data. we believe the actual factor should be closer to the 1.10 theoretically predicted in (11). We fee: that other influences. such as mortar cube strength, air content, and friction between mortar and block. contributed to the large differences between data from the two load conditions reported in (11) and (12).

For this reason in the analysis that follows, we have scaled the strengths from the quarter point tests by 1.10 to relate their values to rest /lts from the other tests, which we believe result in moment distributions more closely representing that in a real wall in a structure subjected to seismic loading.,

The data from references 3.4 and 5 have cube strengths reported corresponding to each tensile strength. The data from referen e 2. on the other hand, gives only the average cube strength from the 8 walls tested. For this reason, the data from reference 2 has been analysed separately.

3.1 DESCRIPTION

OF STATI3TICAL ANALYST-3 Statistical analyses were carried out for two cases: uniform load data and point load data. The reason for this separation is indicated in the preceding

~

section. The tensile strength data from references 3 and 4 were first scaled by 1.10 to account for the lower average tensile strengths expected from quarter point tests. A plot of the tensile strength normal to the bed joint against the corresponding mortar cube strength was then made for all the data from references 3.4 and 5. This plot is shown in Figure 1. Two least squares fits to this data were then made. The first was of the form 5

Y= kX n and the second was of the forrr Y= k /X wheie Y = tensile strength normal to bed joint X= mortar cube strength The resulting curves are also p!otted in Figure 1. The comparison indicates that the expressions found in coatss with the tensile strength as a function of the square root of the cube strength are reasonable and in close agreement with the optimum fit when the exponent is not constrained to a value of 0.5. In view of the closeness of the two curves, the scatter of the data and the ACI-531 code use of functions involving the square root of the cube strength. the second Curve will be used herein. Accepting this relationship between tensile bond strength and mortar cube strength, all data can then be normalized by dividing the test tensile strength c/

the square root of the corresponding rnortar cube strength. This gives 35 normalized samples for the point load cases and 8 normailzed samples for the uniform load cases. For each group, the following parameters were computed:

(i) Sample mean. I(

(ii) Sample standard deviation (minimum mean square estimator), s These statistics were then used as the parameters for the distribution of the population. For each of the two cases (uniform load data and point load data) two underlying distributions were assumed, and the effect of the choice of distribution on the results was examined. The more reasonable distribution was then aCcapted. The two underlying distributions were the normal distribution and the gamma distribution The 95% confidence interval for the mean of the population m was calculated assuming that the normalized variable:

i( - m S/6 is 1-distributed, and that the actual population standard deviation, c . is unknown.

Here n is the sample size.

For the case of the underlying distribution being normal. confidence intervals on the parameters m-lCT . m-20 and m-30' were estimated from the confidence interval on the mean and the sample standard deviation. For the case of the underlying distribution being gamma, a different approach was taken m-lCT corresponar to a value of the cumulative distribution function equal to 0.1587 for the normal distribJtion. This means that a httle under 16% of the area under the probability censity curve lies to the left of m-1(r . Similarly, m and m-3 cr correspond to values of 0.02275 and 0.00135 on the cumulative 6

i km _--___-__._.__ ________._ -____ __ _ _ _ _ _ _ . _ _ - _ _ __ _ _ _ _ _ _ - _ - _ _ _ _ _ - - . _ _ . _ _ _ _ _ _ _ _ _ _ _ _ _ _ - - . _ _ _ _ _ _ _ . . m.____..m_.__m_ _ _ _ _ . _ . ___.-______.__-__-___.a_______.-____m____

distribution function respectively. Based on the confidence interval fe 'he mean, confidence intervals were calculated for values of the gamma distribution for which its cumulative distribution function had values of 0.1587. 0.02275 and 0.00135 respectively.

These actual distributions were then compared with the criteria specified allowable tensile stress normal to the bed joint. i.e. 0.5 V m for the OBE condition.

and 1.33 times that value for the SSE condition. Probabilities that the criteria specified allowable stress would exceed the actual joint strength based on the test results and scaled to a mortar cube strength of 1800 psi were calculated under two assumptions: firstly, that the population mean was equal to the sample mean, and secondly, tnat it was at ;he lower end of the 95%

confictince interval. These conditions are termed A and a respectively in tne table in Section L!.6.

Finally safety factors Dased on the 95% confidence interval for the mean were calculated.

l 3.2 RESULTS OF STATISTICAL ANALYSES 3.2.1 Sample iltatistics in the table below the test tensile strengths normal to the bed joint' have been normalized by dividing each strength by the square root of the corresponding mortar cube strength.

Normalized Normalized Uniform Load Data Point Load Data Sample Size 8 35 Sample Mean 1.1850 0.9621 J

Sample Standard 0.2778 '

O.3501 Deviation Coefficient of 23% 36%

Variation 3.2.2 Confidence Intervals on the Population Mean The normalized variables analyzed in Section 3.2.1 are transformed to real tensile strengths normal to the bed joint by multiplying the normalized variable by the square root of the actual cube strength of interest.

In the case of the Point Beach Nuclear Power Plant the appropriate Cube strength is 1800 psi.

, 7 l

)

The following confidence intervals result:

Uniform Load Data 40.4 psi < M < 60.1 psi Point Load Data 35.7 psi < M < 45.9 psi 3.2.3 Discussion of Normal vs Gamma Distribution The normal distribution is well known, and requires no discussion other than the fact that it is a symmetric distribution with possible values in the range (-oo. 00) We are concerned in this study with data that can only assume positive values (tensile strength). and this is a possible problem with using the normal distribution. For the case of the normal distribution fitted to the point load, normalized test data, approximately 1% of the total area under the probability density curve lies in the range of negative values. This will lead to erroneously high probabilities of exceedence. The Gamma distribution on the other hand, cannot assume negative values. and its shape may be adjusted by varying the parameters k and A.

f x , A(Ax) e X?.0 X (k-1)!

The distribution has a mean value of k / A and a coefficient of variation of 1/ Vii. Thus the value of k is adjusted to give the coefficient of variation observed from the sample, and then \ is colculated to give the correct mean value. The following values of k and 1 arise:

Case k A Uniform Load Data 18 15.190 Point Load Data 7 7.276 For large k (> l 5). the gamma distribution and the normal distribution are extremely close. Thus the normal distribution is used for the uniform load data, and the gamma distribution is used for the point load data.

The histogram of the normalized point load data is shown in Figure

2. and the gamma distribution which best fits the data is also shown.

It should be noted that there is no physical reason why tensile strengths normal to the bed joint should have any particular distribution. However, the gamma distribution can assume a wide vatlety of shapes by varying the parameters k and A. We have chosen the gamma distribution for the point load data because it describes the test data far more accurately than does the normal distributloa. It should also be noted that any distribution fitted to a relative?y small sample size (35 in this case) can be expected to differ somewhat from the histogram of test date i

8

3.2.4 95% Confidence Levels Corresponding to 10 .20 .and 30 Levels (1) at cumulative distribution function = 0.1587 (10' level)

Uniform Load Data 28.7 psi < X < 48.3 psi Point Load Data 21.6 psi < X < 31.6 psi (ii) at cumulative distribution function = 0.02275 (20- level)

, Uniform Load Data 16.9 psi < X < 36.5 psi i Point Load Data 12.8 pst < X < 21.7 psi (iii) at cumulative distribution function = 0.00135 (30' leven Uniform Load Data 5.1 psi < X < 24.7 psi Point Load Data 7.0 psi < X < 14.2 psi These intervals are displayed graphically in Figure 3.

3.2.5 Safety Factors Based on the Mean The reevaluation criteria specifies that the cube strength for type N mortar shall be limited to m = 750 psi. This leads to " allowable

• tensile stresses normal to the bed jofnt (1/2 Y m ) of 13.69 psi for the OBE condition and 18.21 psi for the SSE conditi8n.

Using the above values for th6 OBE and SSE conditions, and the 95%

confidence interval for the mean strength from the tests. Scaled to a cube strength of 1800 psi, the following limits arise for the safety factor based on the mean.

OBE SSE Uniform Load Data 2.95 < SF < 4.39 2.22 < SF < 3.30 Point Load Data 2.61 < SF < 3.35 1.96 s SF < 2.52 3.2.6 Probabilities of Exceedance The probabilities that the code specified allo'*" ' stress will exceed the available strength based on the test results are as follows:

9

I i

Normal C:;tribution Gamma Distribution Case Key OBE SSE OBE SSE Uniform Load A 0.0010 0.0033 0.0010 0.0033 8 0.1190 0.0301 0.0119 0.0301 Point Load A 0.0336 0.0643 0.0105 0.0401 B 0.0694 ,

0.0119 0.0300 0.0902 d.9.tgL (1) A gives the probabilities of exceedance assuming the populaton mean equals the sample mean. B gives the probabilities of exceedance assuming the popu'stion mean is at the lower end of the 95% confidence interval.

(2) The normal distribution gives satisfactory results for the uniform loao data. However.

more weight should be given to the probabilities calculated from the gamma distribution for the point load case.

3.3 DISCUSSION OF RESULTS The key results for the confidence intervals are plotted in Figure 3. together with the OBE and SSE stresses from the re-evaluation criteria. The width of the confidence intervals is greater for the uniform load data, reflecting the smaller sarrpie size for this data. It is seen that both tha OBE and SSE stresses llo within the "mean minus two standard deviations

• confidence interval from the point load dr tr.. and at the extreme lower end of that same interval for the uniform load data. This is consistent with the values for probabilities that the re-evaluation criteria stresses will exceed the actual tensile strength presented in section 3.2.6.

It can be stated that criteria specified allowable stresses will exceed the at,tual tensile strength of the mortar normal to the bed joint between 1 and 10 times in 1.000 for OBE events and betwaen 3 and 40 times in 1.000 for SSE events if the population mean strength is taken at the center of the 95% confidence interval. If one considers the extreme case where the population mean is taken to be at the lower end of its 95% confidence interval, then thase figures become between 12 and 30 times in 1.000 for OBE events (<3%) and between 3 and 9 times in 100 for SSE events (<9%).

Given the extreme nature of the assumption on which these second estimates are based, these probabil!!ies of exceedance are deemed satisfactory.

Alternatively, instead of calculating probabilities of exceedance, one may take 10

the same data, cad calculate f actors of safety based on the me9n. If this is done for the OBE events. using the full range of the 95% cc 6dence interval for the population mean. and taking the extremes from uniform and point load data. the safety factor lies in the re je 2.61 < SF < 4.39.

Similarly, for SSE events, the range is 1.96 < SF < 3.30.

. The point load data must be viewed as a lower bound on the safety fu,ior and an upper bound on the probability of exceedence, since 14 of its 35 sample points are from tests carried out on test specimens cured for only 15 days. We have not attemplad to disguise this data, but its effect on the results (lower and upr'er bounds as discussed above) must be realized.

l 11

4 EVALUATION OF SHAKE TABLE TEST RESIM IS As stated in Section 2.1. the shake table test results of unreinforced masonry walls cannot be used to determine their tensile strength. However they are valuable in assessing the out-of-plane pcst-cracked performance of the walls.

After an unreinforced wall was cracked in a given test. it was ':apable of resisting nigher additional seismic loading. The moments to which the wall was subjected during these tests was Calculated. In most instances, in References 7 and 8.

These moments were based on the assumption that the walls were pinned at the roof connection and pinned at their base.

The apparent strength associated with the cracked state of the unreinforced walls can be estimated using the preceding assumptions on boundary conditions to determine the moments to which the wall is subjected. The ratio of apparent to allowabic strength is determined by dividing the moment to which the cracked wall is subjected by tne moment capacity based on the allowat4s stress for an SSE event.

The pertinent results of each house test have been extracted from References 7 and 8 and summarized in the following subsections. The ratio of apparent to allowable strength of the cracked wall is also given for each test that was performed after the crack occurred. This is based on an allowahle moment capacity in each test. The allowable tensile stress for an SSE event given in the reevaluation criteria is 18.2 psi.

4.1 RESULTS OF TESTS ON HOUSE 1 The unreinforced out-of plane masonry wall (W4) in House 1. shown in Fig.

3.1 of Reference 7. was 8 ft. 8 inches high by 8 ft. long constructed from 6-inch-wide hollow concrete block units. The wall cracked. when it was non-load bearing, along a horizontal joint 2 ft. from the top during Test No. 20 (El Centro 0.21g) It was repaired with a surface bonding material after this test and cracked at the same location during Test No. 27. Paccima

- 0.499 In this cracked state the wall was capable of adequately resisting two additional tests, acolma - 0.63g and El Centro 0.59g.

The moment capacity of the wall based on an allowable stress of 18.2 psi and a face shell thickness of 1.37 inches is 994 lb-in./f t. The ratio of apparent to allowable st.'ength of the cracked wall is given in the following tabulation.

Earthquake Moment on Ratio of Test No. and g-level Cracked Wall Apparent to Allowable (Ib-in/ft) Strength 28 Pacolma (0.63g) 2560 2.57 29 El Centro (0.59g) 2214 2.23 f

12

4.2 RcSULTS OF TESTS ON HOUSE 2 The unrel, forced out-of-plane masonry wall (A1) in House 2 shown in Fig.

3.3 of Refervoce / was 8 feet 8 inches IMan by 14 ft. long with a window and 'Joor opening. The wall was constructed from 6-inch-wide hollow concrete block units. The wall cracked when it was non-load bearing, above orizontal joint at the bottom of the window opening during Test No

4. El Centro - 0.33g. In this crac,ked state the wall was capable of adeqtWly resisting five additional tests. El Centro - 0.45g. Taft - 0.40g and Pacolma - 0.279 0.39g and 0.51g.

The moment capacity of the wall based on an allow 9ble stress of 18.2 psi and a face shell thickness of 1.12 inches is 8Sc lb-in/ft. The ratio of apparent to allowable strength of the cracked state of the wall is given in the following tabulation.

Earthquake Moment on Ratio of Test No. and g-level Cracked Wall Apparent to Allowable (ib-in/f t) Strength 16 El Centro (0.45g) 1325 1.48 16 Taft (0.40g) 1065 1.19 17 Pacolma (0.27g) 977 1.09 18 Pacoima (0.39g) 1518 1.70 19 Pacolma (0.51g) 2364 2.64 4.3 RESULTS OF TESTS ON HOUSE 4 The unreinforced out-of-plane piers (Nos.1 and 3) shown in Fig. 3.3 of Reference 8, were 8 ft. 8 inches high by 3 ft. 4 inches wide. They were constructed from 6-inch-wide hollow concrete block units. Pier 1 was cracked 4 ft. 8 inches from the base before testing began. Pier 3 cracked 7 ft.

4 inches from the base during Test No. 2. In the cracked state Pier 1 was capable of resisting Test Nos. 1. 2. 3. and 4 and Pier 3 resisted Test Nos. 3 and 4. Test No. 2 was a two component El Centro earthquake witn a peak horizontal component of 0.34g end a peak vertical component M 0.26g. Test No. 3 was a two component Taft earthquake with a peak horizontal Component of 0.299 and a peek vertical component of 0.22g.

Test No.4 was a single horizontal component Pacolma earthquake with a peak g-level of 0.32g.

Unfortunately. the accelerometers were removed from Piers 1 and 3 after Test No. 3. so moments on tne piers could not be calculated for Test No. 4.

The moment capacity of the piers based on an allowable stress of 18.2 psi and a face shell thickness rf 1.02 inches is 852 lb-in/ft. The ratio of apparent to allowable strength M the cracked state of the piers is given in the following tabulation.

13

Moment on Ratio of Pier Test Earthquake Cracked Wall Apparent to Allowable No. and g-level (Ib-in/f t) Strength 1 2 El Centro (0.34/0.21g) 1316 1.54 1 3 Taft (0.39/0.22g) 1372 1.61 3 3 Taft (0.29/0.22g) 560 0.66 4.4 OlSCUSSION OF TEST RESULTS The shaking table test results provide the opportunity to assess the out-of-plane performance of cracked concrete block masonry walls subjected to seismic loads. The assessment was made with the aid of a ratio of apparent to allowable strength based on the mume,nts that the wall was subjected to in the cracked state. The allowable strength was based on the allowable stress for an SSE event.

In House 1 two tests were performed on the cracked out-of- plane walls and the maximum ratio of apparent to allowable strength was 2.57. For House 2 five tests were performed on the cracked out-of plane walls and the maximum ratio of apparent to allowable strength was 2.64. For House 4 five tests were performed on the cracked out-of-plane piers, however data is only availablE for threo of the tests because the instrumentation was removed from the piers for one of the tests. The maximum ratic of apparent to allowable strength for the three tests was 1.61. It should be noted that the maximum ratio of apparent to allowable strength catermined in each of the three House tests is a lower bound for each test because test'ag was stopped before the limit state of each of the cracked out-of-plane walls was determined.

This part of the study is more qualitaHve than quantitative, and thus ,t is not significant thet these shaking taolf tests were carried out on walls constructed from type S mortar, whereas =/pe N was specified for the Point Beach Nuclear Power Plant. It is obvious that tho post-cracking strength is not effected by the mortar type.

Based on the preceding results, it is apparent that a significant ratio of apparent to allowable strength exists for the out-of-plane walls when they are subjected to seismic loads in the cracked state.

l

[

14

5 SPECIFIC RESPONCE IO ACTION ITEM 4 The preceding sections provide background information and a s,ummary of the te s data required to respond to each of the five parts of Action item 4. The response to each pdnt follows:

A. Applicablity of NCMA and other test data to the Point Beach masonry walls, with particular emphasis on the static nature of the test as compared to the dynamic loading being evaluated.

The applicability of the static. monotonic tests is given in detail in Section 2.1 and can be summarized as follows:

1) An unreinforced wall resLonds elastically to seismic loads uatii it cracks.

2) 'here is no evidence that Indicates ere is a degradation of the tensile strength normal to the bed joint associated with seismic loads.

3) Only tests that used Type N mortar specified by proportion in ASTM C270 were used to evaluate the margins of safety associated with the allowable stressm.

B. Discussion of the results obtained from dynamic tests conducted at Berkeley and discuss their nonapplicability with respect to the bed joint tensile stresses.

The sisaking table tests performed at Berkeley cannot be relied upon to determine the bed joint tensile stresses because of the difficulties involved in assessing the cracked state of the unreinforced walls before the tests began.

As dise".ssed in detail in Sectionc 2 and 2.1. It was clear that the walls were cracked prior to testing, however the extent of cracking was difficult to assess. The major contribution of the Berkeley tests was the information they provided on the post-cracked performance of unreinforced c Jt-of-plane walls subjected to seismic loads.

C. Discussion and justification for higher allowable tensile stresses for SSE than those allowed for OBE. It is the NRC position that such increase should not e 9eed 33-1/3% of OBE i values.

A higher allowable tensile stress proposed for SSE seismic loads over that allowed for OBE loads for the following reasons.

15

1) The probability of an SSE event is considerably lower than that of OBE event and as a consequence a lower ratio of apparent to allowable strength is acceptatie.

2) Jf a lower ratio of apparent to allowable strength is accepted for an SSE event, there is still a considerable margin of safety after first significant cracking has occurred.

This is discussed quantitatively in the next item.

3) Traditionally higher allowable stresses

'iave been allowed for SSE events and there does not appear to be any justification for not permitting this for the tensile stress ncimal to the bed joint.

D. Technical discussion d? r.onstrating that high safety margins in wall capacity exist eve 4 ster first s!g ni'icant cracking has occurred due to out-of-plane shaking ' light of the tests conducted at Berkeley. Also, provi . . cuantitative safety margins obtained by dividing experimental ultimate wali capacities with the wall capacity based on allowable stresses.

A technical discussion of the tests conducted at Berkeley is given ' Section 4. Lower bounds on the quantitative safety mar 9;..s. after cracking has occurred. for thrae different tests are 2.57, 2.74 and 1.61. These facW; of safe'y are based on an allowable tenwie stress !cr an SSE event of 18.2 psi which has a one-third inc ease over that permitted for an OBE event.

E. Statistical summary of the static monotonic test data as described in item 6.a.

The presentation and discussion of the statistJcal summary of the static, monotonic test data is giver' in Section 3.

It is based on 8 uniform loao tests and 55 point load tests.

The 95% confidence intervals of tensile stresses normal to the bec joint f'>r the uniform load test data and point load test data arc presented in sections 3.2.2 and 3.2.4 and are displayed graphically in Figure 3.

Remembering that the point loao data will give an upper 16

bound on the exceedance probability due to the inclusion of a significant amount of 15-day data, and taking the case where the population mean is equa! to the sample mean. the probability taat the a.Instable tenslia stress of 13.7 psi for an OBE event and 18.2 psi for an SSE event will exceed the tensile strength of the mortar normal to the bed joint obtained 'from the tests will be less than once in 100 times for an OBE event and less than 4 times in 100 for an SSE event.

The range of the factor of safety against the allowable stress of 13.7 psi for an OBE event exceeding the actual tensile strength is 2.(1 to 4.39. The corresponding range for the allowable stre ,s of 18.2 psi for an SSE event is 1.96 to 3.30 17 h

6 CONCLUSIONS The purpose of Action item 4 was to provide a detailed analysis of the available test data to justify the allowable tensile stresses normal to the bed joint used in the reevaluation criteria. The values specified in the revised criteria are 0.5 V m for an OBE event and 0.67 N for g an SSE event. Action item 1 specifica91y limits the strength of the mortar to 750 psi. despite the fact that test samples taken from the walls in the plant indicate that the mortar strength is in excess of 1800 psi. Therefore. In reality. the allowable tensile stresses normal to the bed joint are limited to 13.7 psi for an OBE event and 18.2 psi for an SSE event. In view of the statistical analysis presented herein, this cut-off at 750 psi for the mortar strength is reasonable.

The range of the factors of safety, based on the test data and scaled to a cube strength of 1800 psi, is 2.61 to 4.39 for an OBE event and 1.96 to 3.30 for an SSE event. These factors of safety are based on the 95% confidence intervalt af the mean strength of the test data. In addition to these factors of safetj, the shaking table tests performed 21 Berkeley indicate that there is a significant additional ratio of apparent to allowable strength once the walls have cracked and are subjected to seismic loads. It is therefore concluded that the allowable tersile stresses normal to the bed joint of 13.7 psi and 18.2 pst for OBE a'.d SSE events respectively are reasonable values to use in the *eavaluation criteria for the Point Beach Nuclear Power Plant.

18

7 REFERENCES

1. Copeland. H.E.. and Saxer. E.L. " Tests of Structural Bond of Masonry Mortars to Concrete Block". Proceedings. American Concrete irsstitute.

Vol. 61. No. 11. Nov.. 1964.

2. Hedstrom. R.O.. " Load Tests of Patterned Concrete Masonry Walls".

Proceedings. American Concrete Institute. Vol. 57. P.1265,1961.

3. Fishburn. Cyrus C. "Effect of Mortar Properties on Strength of Masonry *,

Monograph 36. National Bureau of Standards.1961.

4. Whittemore. S.L.. Stang. Ambrose H. and Parsons.

• Structural Properties of Six Masonry Wall Constructions". Building Materials and Structures Report No. 5. National Bureau of Standards.1938.

5. Richart. Frank E.. Moorman, Robert B. B., and Woodworth. Paul.

• Strength and Stability of Concrete Masonry Walls". Bulletin No. 251. Engineering Experiment Station, University of Illinois 1932.

6. Unpublished Data. National Concrete Masonry Association.

7. Gulkan. P.. Mayes. R.L., and Clough. R.W., ' Shaking Table Study of Single-Story Masonry Houses. Volume 1: Test Structures 1 and 2".

EERC Report No. 79-23. Sept.. 1979.

8. Gulkan. P., Mayes. R.L. and Clough, R.W., " Shaking Table Study of Single-Story Masonry Houses. Volume 2: Test Stoctures 3 and 4*.

EERC Report No. 79-24 Sept.. 1979.

9. Clough. R.W. . Mayes. R.L., and Gulkan, P., ' Shaking Table Study of Single-Story Masonry Houses. Jolume 3: Summary. Conclusions and Recommendations". EERC Report No. 79-25. Sept.. 1979.

10.

Mayes. R.L.. Omote. Y.. and Clough, R.W., " Cyclic Shear Tests of Masonry Piers. Volume 1 Test Results". EERC Report No. 76-8. May, 1976.

11. Monk. C.B., " Transverse Strengtn of Masonry Walls'. Special Publication No.166. American Society for Testing and Materials 1954.

12.

"Research Data and Comments in Support of: Recommended Building Code Requirements for Engineered Concrete Masonry' National Concrete masonry Association.

19

l l

l 1

f - TENSILE STRENGTH NORMAL TO THE LED JOINT ('s ci) t e m o a m .

O O O O O o

%~ -4 , 1 1 I s N \

. N \

N \

\ \

. N \

\ \

\ \

m -

N \

5 \ \

\

E -

\

m \ g g -

x\

m z

\\

g\

e \g g g a

m o g gs m

-t E 2 O O m m 30

!$ SO \

g n -

O\\ o

<e

o - o \\ O M J O -

\ \ g @O r m g gg g c m 6 i 0

D -I - '

-4 2 m

C '

\

3 Z g . I \ b

@ $ IG g O M n . \ Q m u g \

m E o g

\ C O \

m I i

$ $ _ O 1 i

y 8 5 e i \

l . .4 \- \ a l

a eo (so \

~

g, l 3 m x 1 5 G I i

• 1 t$ \o. #

k {O Q \g

@ e o @ 3

= .O . t \ O) n a s a \

- \

S 1 ou l

i 20

- - - . _ _ . - . . . . - _ _ _ . . , . .. , _ _ _ _ , . , _ _ _ . . _ _ _ , . _. .. ~_ -

i 1

a -

i 12 -

10 -

Gamma Distribution (k = 7,k7.276) 8 -

/' s s a / x c / s i o s a 6- / N g / N

/ N  ;.

n- N l

\  ;

E 4- / g ~

D /  %

O /  %

% / 's %. l 2- / '

d /

/ x 2 ' '-

- , ~ - >

e

)%

0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 3 i Vm' Sample Mean 0.962 FIGURE 2 NORMALIZED POINT LOAD DATA d

Ug o $ 0g .O$R oo "!

c O- - _ - - _

k FI 1

1 0 i\

G 3 O U 7 B ".

R p E E s

\

\

i 3 1 S O

C N _9 8 S 2E p

s 2

0 1

\

i F

I 5

D  %

E R N _C e C _O -

E N e 3 1

_F v 0 I

I N ID a l

T u E _E

_N a t

R C i V E o A n l L I N C S r 4 T i t

0 I

F E e O R r i

R V a P A L

O S P

U O I L 5 I A _N f 0 T

I O

N S

T K A E T

I 6 I Y

S 0  :

T I

C S k

\

f \

(

p P U si o

i n

i

) y n f t o r

L m o

a L d o a

D d a

t D a a t

a 1

7